Abstract
Methane seeps are unique features on the seafloor where methane and reduced fluids permeate the sediment, providing an energy-rich habitat for endemic marine fauna. Extensive authigenic carbonate slabs are a common feature of seeps, and fossil seep carbonates offer a window into the ecology of the ancient seafloor. Recent studies have demonstrated that microorganisms inhabiting these seep carbonates, called endoliths, remain metabolically active even after being encrusted, and a more thorough understanding of the community dynamics of these systems is needed before the ecological history of fossil seeps can be correctly inferred from the rock record. We analyzed the bacterial and archaeal communities of 20 native carbonates from zones of differing activity and 28 carbonates transplanted between those zones at Mound 12, a methane seep on the Pacific active margin of Costa Rica, along with 15 incubations deployed over a period of 16 months to provide context to understand anaerobic methanotroph (ANME) niche differentiation and fossil seep histories. Our results demonstrated the prevalence of the ANME subgroup 1 within native and transplanted Mound 12 carbonates but not within the incubated experiments, indicating that ANME endoliths are the result of self-entombment rather than preferential attachment. We also demonstrated that the response of carbonate endolithic communities to transplantation was complex. The relative proportion of ANME-1 significantly increased in carbonates transplanted to a low-activity zone, whereas the community of samples transplanted to inactive zones did not significantly change. These results provide insight into how seep carbonates are formed and how the communities change over time, providing a framework for interpreting fossil seep biomarkers.
Introduction
Methane seeps are marine regions where methane-rich fluids permeate the seafloor and stand out from the rest of the benthos by virtue of their geochemistry, richness and diversity of life, and abundance of carbonate-rich rocks that vary in size, ranging from microcrystalline precipitates to square kilometer-size pavements (Levin, 2005; Naehr et al., 2007; Joseph, 2017;
Georgieva et al., 2019). Thousands of methane seeps and hydrothermal vents have been identified on the seafloor, often the result of tectonically-driven seawater advection that is re- emitted enriched in dissolved nutrients and gasses (Tivey, 2007; Beaulieu et al., 2013; Beaulieu et al., 2015). The bedrock metabolism of these communities is the anaerobic oxidation of methane (AOM) with sulfate, mediated by a partnership of sulfate-reducing bacteria (SRB) and
anaerobic methanotrophic (ANME) archaea (Boetius et al., 2000; Orphan et al., 2001a; Orphan et al., 2009; Ruff et al., 2015). AOM consumes 50-80% of sediment methane, preventing its escape to the benthic hydrosphere or, ultimately, the atmosphere (Reeburgh, 2007a; Boetius and Wenzhöfer, 2013; Levin et al., 2016). The sulfide that results from AOM is also captured as a chemical substrate by a variety of unique or endemic organisms with sulfide-oxidizing bacterial symbiotes, including species of mussels, clams, crabs, and tubeworms (Levin, 2005; Cordes et al., 2010; Bowden et al., 2013; Goffredi et al., 2014; Levin et al., 2015; Amon et al., 2017; Ashford et al., 2020; Soares Pereira, 2020).
The coupling of sulfate reduction with methane oxidation by this microbial partnership produces alkalinity and promotes the formation of a variety of authigenic carbonate minerals up to millions of cubic feet in volume (Aloisi et al., 2002; Klaucke et al., 2008; Mason et al., 2015).
The carbonates themselves continue to act as a substrate for some ANME, which live within the rock matrix as endoliths (Marlow et al., 2014a; Case et al., 2015; Mason et al., 2015). Although many endoliths gain nutrients and metabolites from their host rock, ANME endoliths are unusual in that the formation of authigenic carbonate presumably restricts the diffusion of methane and sulfate into the rock (Hovland, 2002; Walker and Pace, 2007). Nevertheless, endolithic ANME have been shown to continue to oxidize methane, indicating that the often- vuggy and micritic carbonate matrix is porous enough to permit methane diffusion (Marlow et al., 2014a; Marlow et al., 2014b).
Any organism that embeds itself in rock represents a potential candidate for fossilization and biomarker preservation, and fossil seeps are not uncommon in the geologic record (Blumenberg et al., 2004; Stadnitskaia et al., 2008b; Bailey et al., 2010; Georgieva et al., 2019).
Several archaea-derived lipids, including crocetane, archaeol, sn-2 hydroxyarchaeol, and glycerol dialkyl glycerol tetraethers (GDGT) have been detected in multiple fossil authigenic seep carbonate mounds, and the relative proportions of these lipids can be analyzed to compute the composition of the endolithic community at the time of fossilization (Blumenberg et al., 2004;
Stadnitskaia et al., 2008a; Stadnitskaia et al., 2008b). Methane seeps are dynamic, however, with seepage flux changing or shifting spatially on scales from days to years to centuries (Tryon et al., 1999; Torres et al., 2002). Little data exists that reveals how the endolithic communities of methane seep carbonates respond to these changes, but an understanding of the community dynamics is essential for interpreting the historical record of these seeps. Biomarkers that are
extracted from seep carbonates are not derived from buried, dead organisms but from a community that can continue to grow and change long after the carbonate crust is formed.
Biomarker studies of seeps, particularly those seeking to understand seep evolution, must be undertaken with caution.
Here, we analyze the endolithic communities of carbonate rocks from Mound 12, a methane seep west of Costa Rica, as well as describe the results of a 16-month transplant experiment that mimics changing seep conditions, in order to better understand methane seep endolithic community dynamics. We find that seep endolithic communities are broadly similar regardless of age, but that these communities may subtly shift as the center of methane flux changes.
Materials and Methods Site description
Mound 12 is a 50 m tall, cone-shaped mud volcano on the Costa Rican Pacific active margin (Linke et al., 2005; Mau et al., 2006; Sahling et al., 2008; Cortés, 2016). The region of main activity lies to the southwest of the mound itself, where evidence of current or prior seep activity extends ~300 m radially outward from the base of the cone. No methane ebullition has been observed at Mound 12, but the active region has extensive carbonate platforms colonized by myriad chemosynthetic organisms including the Costa Rican yeti crab, Kiwa puravida (Thurber et al., 2011; Goffredi et al., 2014). Samples were determined, at the time of collection, to fall into one of three categories: active, transition, or inactive. Transition sites were further divided into inner transition and outer transition by reviewing video footage. Active sites were characterized by exposed carbonate platforms with minimal sediment cover, microbial mat covering visible sediment, and abundant chemosynthetic animals, including bathymodiolin mussels, K. puravida, vesicomyid clams, or vestimentiferan tubeworms. Inner transition (IT) sites were characterized as those with a decreased density of bivalves, absence of microbial mat, shell hash of seep bivalves, light sediment cover of visible carbonate, and/or more abundant non-chemosynthetic organisms such as hydroids, galatheid crabs, or shrimp. Outer transition (OT) sites were characterized by mostly buried carbonate, no visible chemosynthetic organisms, and more abundant non-chemosynthetic organisms such as hydroids or corals. Inactive sites had nearly
entirely buried or absent carbonate and few visible eukaryotes outside of brittle stars, anemones, or fish.
Sampling
All samples were collected using Human Occupied Vessel (HOV) Alvin on two R/V Atlantis expeditions: AT37-13 (May-June 2017) and AT42-03 (October-November 2018).
Carbonate rock samples were collected with Alvin into bioboxes made from thick Delrin plastic.
Using a sterile chisel, pieces of each rock were chipped off. Samples for DNA analysis were placed in sterile Whirl-pak bags (Nasco, Fort Atkinson, Wisconsin) and frozen in liquid nitrogen before storage at -80°C. Additional rock samples were fixed in 2% PFA in 5 mL Nalgene containers overnight at 4°C before washing 3 times with 1.5´ PBS the following day and stored at -20°C in 1:1 3´ PBS:100% ethanol. The water sample was taken with a Niskin bottle fitted onto Alvin at the Yettisburgh active site on dive A4985 in 2018. After returning to the surface, the water was filtered through an inline 0.22 µm Sterivexä filter (MilliporeSigma; Burlington, MA) using a Masterflex L/S peristaltic pump (Model 7528-30, Cole-Parmer; Vernon Hills, IL) at 60 rpm. Tygon peristaltic pump tubing (Saint Gobain; Courbevoie, France) used with the pump was acid-washed in 4% HCl prior to the expedition and flushed with ultrapure water in between uses. ~40 mL of sample was also flushed through the tubing before attaching the filter.
Sample Description
A list of all samples analyzed herein, with accompanying coordinates, activity classifications, is found in the supplementary data file. Twenty natural carbonates were collected from Mound 12: 10 from active sites, 7 from the inner transition (IT) and 3 from the outer transition (OT). We also conducted 28 carbonate transplant experiments. Each transplanted carbonate was collected by Alvin and moved to a zone of differing activity on the same dive. In some cases, natural carbonates were collected alongside transplanted rocks to serve as controls.
Six different types of transplants were conducted: active to IT (n = 4), active to OT (n = 5), active to inactive (n = 5), IT to active (n = 4), IT to OT (n = 5), and IT to inactive (n = 5). These names are shortened with arrows in the following sections, e.g. active®IT. Due to the lack of carbonates in OT or inactive sites (by definition), we were unable to find sufficient samples to conduct transplants from those zones. Two additional carbonates were collected on AT37-13 that were part of long-term incubations on the seafloor. These carbonates were originally
collected from the Costa Rica active margin in 2009 on AT15-44, left to dry for two months, and then replaced on the seafloor in 2010 on AT15-59 in Active areas at Mound 12. Metadata from these samples is sparse, so while these samples were similar to other native active rocks in our analysis, we have considered them separately so as not to confound our analysis.
To test mineralogy-based variation of carbonate microbial communities and to test whether endolithic communities were the result of preferential colonization of a mineral or self- entombment, sterile carbonate chips were also incubated on the seafloor in three locations designated active, transition, and inactive. Each incubation package contained four carbonate types, selected to represent the range of common carbonate mineralogy in the environment:
Iceland spar calcite (Ward’s Science #470025-522; Rochester, NY), coarse-grained crystalline dolomite (Ward’s Science #470025-558), crystalline aragonite from Morocco (Jewel Tunnel Imports; Baldwin Park, CA), and a donut-shaped hydrothermal vent carbonate collected at the Eel River Basin (44.66996, -125.09828) by the Monterey Bay Aquarium Research Institute Remotely Operated Vehicle Tiburon dive 870 in 2005. This donut-shaped rock sample had been kept dry since its collection in 2005. Each of the carbonates were cut into ~1 cm2 chips using a water-cooled low-speed diamond saw. The chips were then dried and any surficial reduced carbon removed with a vacuum plasma cleaner (PE-25; Plasma Etch; Carson City, NV) for 20 minutes; surface cleaning was followed by autoclaving for 30 minutes. Four chips of each mineral type were placed into a 1 mm nylon mesh bag along with 2 autoclaved polyurethane sponges (after (Imachi et al., 2011). Each bag was deployed on the seafloor during AT37-13 in 2017 and recovered during AT42-03 in 2018, resulting in a total incubation time of ~16 months.
DNA Extraction
Natural and transplanted rock samples were processed with the aim of extracting DNA from the endolithic community with minimized contribution from the sediment microbial community. The samples were first washed and sonicated at 8W (Branson Sonifier 150D, Branson Ultrasonics Corp.; Danbury CT) to remove external sediment-associated microorganisms following (Mason et al., 2015). Following the washing, samples were powdered with a sterile mortar and pestle and 0.25 – 0.5 g powder was added to the lysis tubes of the PowerSoil DNA extraction kit (Qiagen; Hilden, Germany). We also performed several additional steps to separate cells from the minerals and lyse them. First, the lysis tubes with powdered
sample were flash frozen in liquid nitrogen and thawed in a 25°C water bath. The thawed tubes were then placed in a sonicating water bath for 90 seconds. The C1 detergent was then added and the samples vortexed briefly to mix before incubating the entire sample in a 65°C water bath for 30 minutes. The heated samples were then placed in a bead beater (FastPrep FP120, Thermo Fisher Scientific; Waltham, MA) at 5.5 m/s speed for 45 seconds. The samples were then centrifuged at 10,000 ´ g for 30 s and processed according to the rest of the manufacturer’s instructions. Our contamination controls included no-sample extraction kit controls as well as a sterilized, unincubated calcite chip that was crushed and extracted in the same manner above.
Since seafloor-incubated sterile carbonates were not expected to contain significant endolithic populations after just 16 months, we did not sonicate those samples before crushing the rock. Instead, those samples were washed with filtered seawater during shipboard recovery to remove sediment and frozen. Upon thawing them in lab, they were placed directly into the sterile mortar and pestle, crushed, and then extracted following the lysis and purification steps outlined above.
DNA Sequencing
The V4-V5 region of the 16S rRNA gene was amplified using archaeal/bacterial primers with Illumina (San Diego, CA) adapters on the 5’ end (515F 5’- TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG-
GTGYCAGCMGCCGCGGTAA-3’ and 926R 5’-
GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG-
CCGYCAATTYMTTTRAGTTT-3’). PCR reaction mix was set up in duplicate for each sample with Q5 Hot Start High-Fidelity 2x Master Mix (New England Biolabs, Ipswich, MA, USA) in a 15 µL reaction volume according to manufacturer’s directions with annealing conditions of 54°C for 30 cycles. Template-free controls were also included for each sequencing run. Duplicate PCR samples were then pooled and barcoded with Illumina Nextera XT index 2 primers that include unique 8-bp barcodes (P5 5’-AATGATACGGCGACCACCGAGATCTACAC-
XXXXXXXX-TCGTCGGCAGCGTC-3’ and P7 5’-
CAAGCAGAAGACGGCATACGAGAT-XXXXXXXX-GTCTCGTGGGCTCGG-3’).
Amplification with barcoded primers used Q5 Hot Start PCR mixture but used 2.5 µL of product in 25 µL of total reaction volume, annealed at 66°C, and cycled only 10 times. Products
were purified using Millipore-Sigma (St. Louis, MO, USA) MultiScreen Plate MSNU03010 with vacuum manifold and quantified using ThermoFisher Scientific (Waltham, MA, USA) QuantIT PicoGreen dsDNA Assay Kit P11496 on the BioRad CFX96 Touch Real-Time PCR Detection System. Barcoded samples were combined in equimolar amounts into a single tube and purified with Qiagen PCR Purification Kit 28104 before submission to Laragen (Culver City, CA) for 250 bp paired end sequencing on Illumina’s MiSeq platform with the addition of 15-20% PhiX control library. Raw read files with accompanying controls were submitted to the NCBI Sequence Read Archive under BioProject Accession Number PRJNA623020.
DNA sequence analysis
712 sequenced samples from across the entire set of AT37-13 and AT42-03 expeditions, including sample-free extraction controls and template-free PCR controls, were processed together in QIIME version 1.8.0 (Caporaso et al., 2010) following a recently published protocol (Mason et al., 2015). Raw sequence pairs were joined requiring a 50 bp overlap (with a maximum of 4 mismatches in overlapping sequence). Contigs were then quality-trimmed, with minimum Phred quality score of 30 (QIIME default is 4), and any sequences with unknown base call (“N”) were removed. Contig sequences were then clustered de novo into operational taxonomic units (OTUs) with 99% similarity using UCLUST, and the most abundant sequence was chosen as representative for each de novo OTU (Edgar, 2018). Taxonomic identification for each representative sequence was assigned using UCLUST with the Silva-132 database (Quast et al., 2013), requiring minimum similarity of 90% to assign a taxonomy; 9 of the top 10 hits were required to share a taxonomic assignment to assign that identification to a query. By comparison, the QIIME defaults include 90% similarity, but only 2 of 3 database hits must share taxonomy for positive assignment. Our SILVA database is appended with 1,197 in-house high-quality, methane seep-derived bacterial and archaeal clones. Any sequences with pintail values >75 were removed. The modified SILVA database is available upon request from the corresponding authors. Singleton OTUs were removed first, and then any OTU which occurred less than 10 times across all samples (or about 0.000001% of the remaining sequences, were also removed.
Known contaminants in PCR reagents and extraction kits as determined by analysis of no- template PCR and no-sample extraction controls run with each MiSeq set were then removed.
After all filtering steps, 66,350 OTUs remained. Samples with fewer than 1,000 sequences recovered from these 66,350 OTUs were removed. In Chapter 2, 268 sediments and carbonates
from Mound 12 were considered; in this study, we analyzed 66 total samples: fifty natural and transplanted carbonates from Mound 12, fifteen sterile substrates incubated at Mound 12 (3 bags of 5 substrates each), and one filtered benthic water sample.
We chose not to rarefy our data consistent with several studies indicating that it impedes detection of differential abundance for little apparent benefit (McMurdie and Holmes, 2014;
Weiss et al., 2017). These studies indicate that using a fourth-root transformation of relative abundance values is computationally simple, reasonably effective at detecting true differential abundance, and does not require discarding valid data. Accordingly, we used a fourth-root transformation prior to non-metric multidimensional scaling (NMDS) ordinations and all downstream analyses, conducted in R along with the vegan ecological statistics package (R Core Team, 2014; Oksanen et al., 2018). NMDS constructs de novo ordinations based on ranked pairwise sample dissimilarities (we used the Bray-Curtis dissimilarity metric), and the fit of the ordination is measured by the stress. Stress values below 0.2 are generally considered acceptable, and all ordinations presented here have stress values < 0.2 with parameter trymax = 100. We also conducted analysis of similarity (ANOSIM), similarity percentage (SIMPER), permutational analysis of variance (PERMANOVA) analyses, and homogeneity of dispersion (PERMDISP), also using vegan in R; p-values < 0.05 after 999 permutations (alpha value) were considered significant. OTU vectors were calculated for the NMDS ordination using the vegan envfit command with 999 permutations. In Fig. 3.3, we included only the most significant vectors (p- values = 0.001) that were also indicated as strong contributors to the variance in SIMPER analysis, i.e. the top 50 OTUs that were identified in pairwise analysis of the transplant treatment identifier (active®inactive, IT (native), IT®active, etc.) by contribution to the variance.
Fig. 3.1: Average microbial community of each treatment by relative abundance. 99%-similar OTUs were binned together by shared taxonomic assignment. The top 25 most abundant taxonomy bins for native and transplant carbonates and sterile carbonate incubations are presented here. ANME-1 bins are colored in dark blue tones, ANME- 2 bins are in yellow and orange, sulfur-metabolizing organisms (including SRB and sulfide-oxidizers) are colored in purple tones, methylotrophs and aerobic methanotrophs are in light blue, other heterotrophs are in green, and nitrogen-metabolizing organisms (including ammonium-oxidizers and nitrate-reducers) are in red. Bins with no putative function are colored in gray. Most bins were not shared between the two charts; the three shared bins are marked with
**. All of the native and transplanted samples contained high proportions of ANME-1 and ANME-2 subgroup organisms, but almost no ANME sequences were detected in incubated carbonate chips. Instead, those samples were dominated by methylotrophic organisms (active area) or ammonium-oxidizing Nitrosopumilis (transition and inactive area).
Results
Native carbonates versus sterile incubations
In our analysis of the colonizing microbial communities, the starkest difference between groups of samples lay between native carbonates (including both natural and transplanted
samples) and the sterile carbonate incubations (Fig. 3.1, Fig. 3.4). Native carbonates were characterized by high relative abundances of anaerobic methanotrophic archaea (ANME) from subclades 1 and 2. In every category of native carbonates, ANME organisms comprised >50%
of the total endolithic microbial community; within IT non-transplanted carbonates, ANME- identified OTUs comprised 77.6% of the community, on average (the maximum). By contrast, no ANME subgroup exceeded 1% average relative abundance in any of the sterile carbonate incubations, even when binning OTUs of the same subgroup together (Fig. 3.1). Instead, sterile incubations from the active site were dominated by organisms from the Marine Methylotrophic Group 2 of Gammaproteobacteria, which are aerobic methylotrophs (Ruff et al., 2013), as well as other possibly methylotrophic organisms from the Rhodobacteraceae and Methylophagaceae families. Transition and inactive sterile incubations were dominated by ammonium-oxidizing archaeal genus Nitrosopumilus (Walker et al., 2010; Qin et al., 2017), and also showed abundant putatively aerobic or sulfate-reducing heterotrophs. The transition and inactive carbonates samples were also the solid substrates that were most similar to the microbial community recovered from the Mound 12 benthic water sample, which was dominated by organisms from order Alteromonadales, including Alteromonas, Pseudoalteromonas, and Marinobacter.
Native carbonate differences by activity
Among the natural and transplant carbonates (excluding the two samples deployed in 2010 for lack of metadata), the treatment identifier (e.g. active, IT, OT, active ® IT, etc.) was significant, but did not strongly correlate with the sample variance (ANOSIM R Value = 0.1453, p-value = 0.015). Our NMDS of the samples, however, indicated that the low correlation may have been the result of a difference in response amongst the groups, and we subsequently conducted pairwise ANOSIM between each treatment identifier. Fig. 2 highlights these differences in the NMDS and Table 3.5 contains the p-values for each ANOSIM and dispersion of each group. Among the native, non-transplanted samples, each pairwise combination of active, IT, and OT was significantly different (Fig. 3.2a); the one-way ANOSIM of all native, non-incubated carbonates also indicated significant difference (R Value = 0.2823, p-value = 0.004). Active samples were by far the most diverse group (although also the most deeply sampled), and much of the residual variation lay within that group.